Definition and evaluation of production test validation methods applied to Vx multiphase flow meters
More Info
expand_more
Abstract
Accurate flow rate measurements are very important in production testing. Originally small and insignificant deviations could propagate in subsequent applications where flow rate data is used as input parameter. Vx multiphase flow meters give much more information compared to conventional test separators. This extra data can be used for extra flow rate validation on the measurements, in order to find out whether it is possible to make sense of the data and if it is applicable on the long term and viable for further applications. Firstly, a comprehensive literature review has been done on current existing validation methods which gave a coherent overview. This work was used to define methods to detect measurement issues which can occur during a production test, defined by experts with considerable field experience. An adjusted choke model is presented which could be run in series with a Vx meter, which is able to detect: a drift in the differential pressure sensor; the under- or overestimation of the calculated gas rate or whether the meter is operating within the designed operation envelope or not. The choke model is tested on 96 data points from actual production tests with satisfactory results for fixed choke data. In addition, a methodology is developed to detect whether a production test has become stable or not. The method is translated into an algorithm which can be run on Vx meter output to decide in an automated fashion if a production test can be concluded, yielding representative production data. The method is tested on 700 production tests in order to define the model thresholds, which showed that the majority of the tests could have been concluded earlier. It has been found that by analyzing the statistical properties of the data it is possible to observe flow pattern transition. Furthermore, by representing certain parameters in the frequency domain, slug flow regimes can be detected and the corresponding slug flow periodicity can easily be subtracted from the data.